AI built around human goals.

Every human has an arc.Ascend yours.

Every person starts somewhere different. Arcend understands the goal, identifies the next step, and measures whether it worked.

Scroll
01  ·  Our story

The belief the whole company is built on.

Every human has an arc.

Every person starts in a different place, moves at a different pace, and faces different constraints. Guidance should be built around one person, not optimized for an average. Progress should reflect the person, not the average.

We believe every arc can ascend.

Human potential is not fixed. With the right guidance, the right next action, and the resolve to keep moving, people learn more, achieve more, and become more than they believed possible. Progress rarely arrives all at once. It compounds.

That is why Arcend exists.

AI can now adapt guidance to what a person knows, wants, and struggles with. Its job is to turn intention into action by making the next step clear. It should not replace human judgment, ambition, or potential. It should strengthen them.

So we build systems that make progress provable.

Meaningful progress has to be measurable. We start with education because scores change, deadlines are fixed, and progress can be measured. But we are building a model of how a person improves. That model does not stop at a test.

02  ·  The problem

Foundation models are not built to measure how people grow.

They excel at answering questions and completing tasks. Helping a person improve requires another signal: evidence of what changed after an intervention.

Why goals are hard to reach Human side
  • 01StartThe goal is clear; the next step is not.
  • 02AdjustPlans break down as circumstances change.
  • 03PersistKnowing what to do does not make it easy to keep going.
  • 04VerifyEffort is visible; real progress is not.
Why today's AI falls short Model side
  • 01DataAbundant conversation data; limited evidence of learning outcomes.
  • 02ObjectiveWhen outcomes go unmeasured, satisfaction becomes the default signal.
  • 03CostAsking, waiting, and withholding answers can conflict with immediate helpfulness.
  • 04ResultMore capable models do not, by themselves, add missing outcome data.
03  ·  Why Arcend

Answers and execution are abundant.
Verifiable outcomes are scarce.

An answer is only a starting point. Arcend maintains continuity around a person's goal and verifies whether each step produced progress.

Why this is possible now.2026 · Now
01

Capabilities spread

Explanations and content are becoming easier to produce. Baseline AI capabilities are spreading quickly.

02

Execution matures

AI can now read state, use tools, and work toward goals, making sustained execution increasingly practical.

03

Value shifts to outcomes

Answers alone do not guarantee progress. Value comes from measurable movement toward a goal.

04

Evidence compounds

When interventions are linked to outcomes, they create evidence foundation models often lack.

04  ·  The solution

Turn a measurable goal into a sequence of
verified next steps.

Each cycle is designed to create measurable progress for the learner and useful evidence for the system.

01 · Plan

Map the goal.

Turn the target, deadline, current state, and available time into ranked next steps.

02 · Intervene

Learn, then verify.

Diagnose the gap, teach what is missing, and test transfer in a fresh context.

03 · Update

Improve the path.

Use each verified result to sharpen the forecast and determine what happens next.

05  ·  The engine

Each session generates evidence.
The system learns from it and improves the next session.

Each session links learner state, error cause, intervention, interaction, and transfer outcome. Together, they build evidence about how people reach goals.

CLOSED LOOP State Reads state learner profile Intervention Chooses move next teaching move Outcome Verifies transfer Decision model Updates what works next
01
learner.state

Read the state

Track what the learner knows, why errors occur, and how new evidence changes the profile.

02
intervention.choose

Choose the intervention

Use one teaching approach at a time; record the decision model's choice and exploration separately.

03
transfer.verify

Verify consistently

New context, new numbers, no hints, first attempt only. Every path ends with the same type of transfer item.

04
model.update

Update the decision model

Use the result to update the learner profile and inform which teaching move comes next.

06  ·  One system

Each verified outcome adds evidence.
Better judgment compounds over time.

Foundation models supply knowledge and reasoning. Arcend builds the goal, intervention, and outcome loop around them. Evidence accumulates as students do real work.

COMPOUND System Captures evidence state → outcome Outcomes Provide signal verified progress Decision model Learns what works next
  • 01

    Foundation models are the supply layer

    They provide knowledge, generation, and reasoning. Improvements in that layer can make our loop cheaper to run.

  • 02

    Outcomes are the learning signal

    Each verified result gives the system evidence about what to try next.

  • 03

    Evidence compounds

    More verified outcomes sharpen the system's judgment. Better judgment improves future guidance.

07  ·  The arc

From test preparation
to broader human goals.

We begin with test preparation, where goals, deadlines, and outcomes are clearly defined. Over time, we aim to apply the same approach to other areas of human development.

Phase 1 · now

Goal-guided learning

Turn a test goal into today's next step, beginning with the SAT and AP.

Phase 2

Adaptive learning paths

Use evidence across sessions to anticipate which interventions will help each learner.

Phase 3

Guidance beyond test preparation

Explore how the same verified next-step loop can support goals in work, health, and life.

08  ·  The team

Founders with experience

Alex Li
Co-founder & CEO
Education × Product
  • 15 years in education, including 11 focused on North America
  • Founded StudyX, reaching 17M registered users
  • Built a No. 1 company in an international education niche
Yuru Shao
Co-founder & CTO
AI × Systems
  • Former engineering manager at Pinterest
  • Built security systems for products with 100M DAU
  • PhD in Computer Science, University of Michigan
Sky Shou
Co-founder & CGO
Growth × Monetization
  • 15+ years in global app growth and monetization
  • Early team member and business lead at QuVideo
  • Led VivaVideo to nearly $100M in annual revenue
Join Arcend

Every arc can ascend.
Help us prove it.

Whether you want to reach your own goal, build the system that helps millions reach theirs, or back what comes next, there is a place for you.

Students & families

Start with the goal you already have and discover what verified progress can feel like.

Get early access ↗

Partners

Schools, publishers, and platforms building on the same engine.

Talk to us ↗

Investors

The full thesis, the data asset, and where the loop goes next.

Request the deck ↗
Arcend at a glance
What Arcend is
AI built around human goals. One system that understands the goal, tracks progress, guides every step, and builds measurable skills.
Approach
A closed-loop system that turns goals into verified progress and learns from every outcome.
Founders
Alex Li (CEO), Yuru Shao (CTO), Sky Shou (CGO).
Contact
[email protected]. Partnerships and press are welcome.